What do you lose by going HDMI-to-DVI through the 990? I can't think of any: the video signal is unaltered, and the audio can even pass through to the display if you use dual-link DVI adapters (based on several user reports).

Let's ask a different question: if Outlaw had elected to hold off on the 990 until they could do HDMI for both audio and video, what would that have meant? They might have been able to do HDMI v1.1 with a delay of maybe a year - possibly more, as it would have forced some pretty significant changes to the existing Sherwood hardware platform. That would have given them HDMI audio support, but the lack of v1.2/v1.3 is widely perceived as "outdated" (was even seen that way before the v1.3 spec was published). Outlaw surely recognized that market perception. If they'd chosen to waith for HDMI v1.3 then we'd very possibly still be waiting for the 990 (over three and a half years after the 990 started shipping). After all, the 990 uses a Cirrus 49400 and moving to the 49700 would have been a logical progression, but the 49700 chip was delayed at least a year and a half. All of a sudden, instead of a May 2005 release with DVI the 990 might have been a fall 2006 release with HDMI v1.1 or a fall 2008 (or later) release with HDMI v1.3. From that viewpoint, the 990's use of DVI continues to make some sense.

The other aspect of "future-proofing" is the use of DVI rather than HDMI video switching (which I still say rather adamently should be licensed separately from full HDMI implementations with a separate logo somehow - maybe "HDMI Video"). The big problem with using HDMI ports for switching is confusion. I've lost count of the number of receivers or processors that I've had to pick through a user manual or run a series of Google searches to determine if the HDMI ports support audio or not. Even a month ago, I ran into a co-worker who thought that all HDMI inputs support audio and video - but manufacturers are still (in 2009) producing entry-level receivers that have HDMI inputs that can't handle audio of any kind. The use of DVI instead of HDMI may create some concerns about DVI / HDMI compatibility for potential 990 owners, but it also avoids the problem of customers buying a 990 under the assumption that it will do HDMI audio.
The market for surround processors is finally going to calm down a little now that HDMI v1.3 is in place. The processor generation that the 997 will be part of includes HDMI v1.3, room correction, and internal video processing. What will come after that? I honestly don't know - there are likely to be some refinements in those technologies, but I don't see anything looming close on the horizon to add significantly to the equation. There will be something new eventually - but there always is, and if you wait to buy until that next big thing is available you'll be watching TV on a 19" analog CRT and listening through the one speaker built into that TV. Personally, I think it'll be a bit longer before we see the next major "new thing" for surround processors.

Network media devices are a completely different matter. There are a lot of different options out there, many competing, and a scarcity of dominant standards. You've got SlimDevice's Squeezebox, Sonos, Roku Labs SoundBridge, and other devices for audio only. You've also got AppleTV, game consoles (XBox360 and PS3), and a number of other devices that can handle audio and video. These can range from a couple hundred dollars to upward of $10,000 (if you mix in something like Kaliedescape). Then there are DVR's (either standalone or built into cable and satellite receivers). Then there are devices that let you download video content - Roku's Netflix box, Amazon Unbox, Vudu, and the like. It's a very active market, but there's a lot of turbulence and fluctuation. It's likely that people who get serious about it may use several different devices for different things. That's the biggest reason that I think letting a media client stay separate from the surround processor is the smartest choice. Let it be a source - it deserves to be. If media clients become as simple, established, and unchanging as AM/FM tuners, then maybe they should be integrated into the surround processor. Even then, though, I don't see it as a necessary inclusion.

Should Outlaw get into that market with a product? I don't know how easy that would be for a company structured the way they are - there's a lot of risk if the horse you invest R&D effort into loses. On the other hand, there have been some suggestions in this thread that Peter Tribeman's upcoming January announcement might be a partnership to sell something like this - although there are plenty of guesses there that would go in a different direction, so it's hardly guaranteed.